Anda di halaman 1dari 37

VI. Relaxations in Polymers and the Glass Transition.

Glass
Transition region
Traditional presentation of relaxation map in
polymers includes 4 regions:
Log G(t)

Segmental
Relaxation -Glass-like behavior;;
-Transition region;
Rubber -Rubbery plateau;
Rouse
Modes Flow -Terminal relaxation (flow).
Reptation

0 1 2 3 4 5 6 7 8 9
10 10 10 10 10 10 10 10 10 10

Log t

Secondary relaxations (
)
Segmental relaxation
(-relaxation) However, there are a few relaxation processes
Fast
even in a glassy state (or at time scales shorter
G(t)

picosecond
Log G

relaxation th segmental
than t l relaxation).
l ti )
In particular, secondary relaxations influence
mechanical properties of polymers.

0 1 2 3 4 5 6 7 8 9 10 11 12 13
10 10 10 10 10 10 10 10 10 10 10 10 10 10

Log t
Frequency map of polymer dynamics
Chain dynamics:
y Rouse and terminal relaxation
Structural relaxation, -process
Vibrations
Secondary relaxations Fast relaxation FREQUENCY,

10-2 1 102 104 106 108 1010 1012 1014 Hz


Mechanical relaxation G
Dielectric Spectroscopy, *() Quasi-optics,TDS

Traditional dielectric spectroscopy IR-spectr.

Light Scattering, Iijj(Q,) Interferometry

Photon Correlation Spectroscopy Raman spectroscopy

Spin-Echo
Neutron Scattering, S(Q,) Back-sc. Time-of-Flight

Inelastic X-ray Scattering, S(Q,) High-Resolution


IXS

Scattering techniques have an advantage due to additional variable wave-vector Q


Time-Temperature Equivalence (Superposition)

Secondary relaxations ()
Segmental relaxation There are many relaxation processes in polymeric
(-relaxation) systems. They appear with different characteristic
Fast times ,
relaxation times, or frequencies.
frequencies The names of
Log G(t)

picosecond
the processes (-, -, -) reflects the order in
relaxation
which the processes appear (excluding chain
relaxation). There is no physical meaning behind
these names. However the -process
names However, process is usually
assigned to the segmental relaxation.
0 1 2 3 4 5 6 7 8 9 10 11 12 13
10 10 10 10 10 10 10 10 10 10 10 10 10 10

Log t
Different experimental techniques should be used
S e c o n d a r y r e la x a t io n s ( ) for measurements of the relaxation spectrum in so
S e g m e n t a l r e la x a t io n
( - r e la x a t io n ) broad time or frequency range. Another way can
g G ()

be to change temperature and measure at fixed


1

frequency. In particular, crossing the segmental


Log G, log

relaxation usually marks the glass transition


temperature Tg (depends on frequency).
T g
Comparison of two plots demonstrates that
T [K ]
variations in time and in temperature can be
equivalent.
Time-temperature equivalence assumes that the
viscoelastic behavior at one temperature can be related to
that at another temperature by a change in the time scale
only.

The compliances at T1 and T2 can be superimposed


exactly by a horizontal shift LogaT. Similarly, results of
dynamic mechanical experiments measured at two
different T can be shifted. aT is the shift factor.

There are different corrections to the simple shift factor.


In particular, superposition should incorporate a small
vertical shift factor: T00/T, is the density.

Molecular rate process with a constant activation energy


Lets assume that there are two different conformational states with energygy
difference , separated by energy barrier V. In that case the relaxation time:
V V
0 exp sec h 0 exp
kT 2kT kT
Here 0~10-12-10-13 s.s One can also introduce a relaxation rate:
1 V 1
0 exp ; 0
kT 0
This gives an activation or Arrhenius temperature
0.01 dependence. It appears as a straight line for ln or ln vs 1/T.
~exp(-V/kT)
exp( V/kT)
log

The slope gives an estimate of the barrier height V:


1E-3
(ln )
V k
1
( )
1E-4 T
k is the Boltzman constant. Comparison of frequencies
3 4 5 6 7 8 9 10

1000/T [1/K] measured at two different temperatures (for example,


Arrhenius temperature dependence frequency of tan maximum) gives the shift factor:
1 V 1 1 T1 T2
log 1 log aT C
2 ln 10 k T2 T1 T1T2

Secondary
S d relaxations
l i usually
ll show
h A h i temperature dependence
Arrhenius d d with
i h an activation
i i energy
V~20 kJ/mol. Segmental relaxation has strongly non-Arrhenius temperature dependence, i.e. its
apparent activation energy is a function of temperature.

In reality there are many corrections to so simple picture. It was found that the shift factor for various
polymeric systems is usually well described using Williams-Landel-Ferry (WLF) equation:
C1 (T TS )
logg a T
C 2 (T TS )
An example of storage compliance
data for poly-n-octyl methacrylate.
Composite curve obtained by
plotting the data with suitable shift
factors Original data were extended
factors.
over 2 decades only. The combined
curve covers 11 orders in frequency.

Temperature dependence of the shift factor obtained for poly-n-octyl


methacrylate in order to obtain the master curve shown on the previous
slide. The temperature dependence is strongly non-Arrhenius and can be
described by the WLF equation.

The time-temperature equivalence (or superposition) is used by many researchers for analysis of
viscoelastic properties of polymers. You find it in all textbooks. It explicitly assumes that all
viscoelastic processes have the same temperature variations. One should keep in mind, however, that in
most cases the time-temperature superposition breaks down. It will be discussed later.
Segmental Relaxation -process
The main
Th i structural
t t l relaxation
l ti ini polymers
l i the
is th segmental, ll d -relaxation.
t l so-called l ti It controls
t l diffusion
diff i
and viscosity, rotation of monomers. The -process is the relaxation on time scale shorter than the
rubbery plateau and the Rouse modes. It is responsible for the transition region and is associated with
the glass transition.

It is usually ascribed to micro-Brownian motion of chain segments. Most authors agree that the -
process is related to conformational changes (like gauche trans transition). Because the glass transition
is directly related to the -relaxation, both, Tg and segmental relaxation (the -process), show the same
dependence on Mw and crosslinking.
crosslinking

Many similarities between the glass transition in low-weight


molecular systems and in polymers suggest that the chain
connectivity is not required for this process.
process

A correlation between the dielectric relaxation time s for the


local segmental mode in dilute solution and Tg [Dielectric
Spectroscopy of Polymeric Materials, Eds. J.Runt, J.Fitzgeraldp.279]. This
observation supports direct relationship between segmental
relaxation of a chain and Tg.
Spectral Shape of a Relaxation Peak
A simple
p relaxation pprocess usually
y has a single
g exponential
p relaxation, G(t) p( ).
( )exp(-t/
G"()~

Distribution of
1.0

0.8

Single exponential relaxation


G(t)~exp(-t/)

og G"()
G(t)

H()
0.6
-1
~
appears as a Lorentzian in the
~
0.4 frequency
q y domain and

Lo
1/
1/e

0.2
corresponds to a single

0.0

Time Log Log

The -process is not a single exponential decay,


decay it is usually well described by a stretched exponential
relaxation, the so-called Kolrausch-Williams-Watts (KWW) equation: t KWW
G (t ) exp

1.0

0.8

-b
Log G"()

G(t)~exp(-(t/) )

~ H() is a distribution
G(t)

H()
~
0.6

0.4
1/e
of , i.e. it
characterizes how

0.2

00
0.0
many relaxators have
Log
Time Log relaxation time .

Stretched exponential process is usually approximated by a Cole-Davidson distribution function. It is


asymmetric and has extended high-frequency tail. Generally, bKWW.
Thi stretched-exponential
This t t h d ti l shape
h i nott specific
is ifi for th -relaxation
f the l ti in i polymers
l only.
l Many
M l
low-
molecular weight liquids (especially in the supercooled state) demonstrate that. However, polymeric
systems usually have higher stretching (e.g. lower values of KWW), KWW~0.35-0.7.
Intermediate scattering function (t) in PPG
[Bergman,
Bergman et al.
al Phys.Rev.B 56 11619 (1997)] shows
Phys Rev B 56,
strong stretching of segmental relaxation.
Dielectric relaxation spectra in PDMS are also
stretched [Hintermeyer, et al. Macromolecules 41, 9335(2008)].
The shape parameter depends on the system and varies with
temperature. However, at higher T the temperature variations of
KWW is weak.

There are two basic reasons for the stretched spectrum:


1. The process corresponds to a weighted sum of elementary
g
processes each having a correlation function that may be
Stretching parameter for different polymers exponential
i l in
i time.
i E.g.
E it i may be
b equivalent
i l to a
[Colmenero, et al. JPCM 11, A363 (1999)]. distribution of relaxation times
2. The process has a natural non-exponential dependence.
Microscopic Picture of Segmental Relaxation
Microscopic mechanism of the segmental relaxation remains unclear. Mechanical, dielectric
relaxation and light scattering measurements do not provide microscopic details of the motion. NMR
and neutron scattering can provide this information.

Example of dielectric relaxation spectra in PB [Arbe,


Arbe et al.,
al Phys.Rev.B 54 3853 (1996)].
Phys Rev B 54, ] Two relaxation processes (- and
-) with strongly different temperature dependence are observed. provides no microscopic information.

Neutron spectroscopy probes atomic motion on an inter- and


intra-molecular scales.
Static structure factor, S(Q), in PB. The peak in at Q~1.5 -1
corresponds to inter-chain distance, while peak at Q~3-1
corresponds to intra-chain distance in PB.
PB
Neutron spin-echo (NSE) measures Intermediate scattering
function S(Q,t) directly.
NSE measurements have been done on deuterated PB [Arbe,
et al.,
al Phys.Rev.B 54 3853 (1996)].
Phys Rev B 54, ] The measurements do not
cover sufficiently broad time range. The authors used a
shift factor ~ (the viscosity time scale) and received a
master curve for Q~1.5 -1 (the fist peak, corresponds to
intermolecular distance) but no master curve is observed
for Q~2.7 -1. The process at Q~1.5 -1 was interpreted as
the -process and at Q~2.7 -1 as the -process.

Decay of correlation function at the inter-molecular


inter molecular Q with
characteristic relaxation time ~ suggests that the
segmental relaxation corresponds to a motion of a chain
relative to its neighbors (decay of correlations between
neighbor chains).
chains)

In contrast, the -process in PB has rather intra-molecular


origin.
Incoherent scattering (self-correlation) shows strong dependence of S(Q,t) on Q. In particular,
characteristic relaxation time of segmental relaxation has strong Q-dependence, even stronger
th ~Q
than Q2. These
Th result
lt suggestt that
th t the
th segmental
t l relaxation
l ti is i a diffusive-like
diff i lik motion.
ti

Single chain S(Q,t) (left)


and average measured at
two different temperatures
in PIB (right) [Richter, et al.
J.Phys.Cond.Matt. 11,, A285
(1999)].
Open symbols show
at T
T=390K
390K shifted with
rheological shift factor.

Example of analysis of (Q) dependence in PVME


[C l
[Colmenero, ett al.
l J.Phys.Cond.Matt.
J Ph C d M tt 11, (1999)] does not show
11 A363 (1999)].
normal behavior ~Q , it varies much stronger ~Q-2/. The
-2

conclusion is that this is a diffusion-like motion, but


strongly stretched (not a trivial Brownian motion).
Deviations at higher Q (Q>2 A-11) are explained using jump
model (elementary jumps) and an estimate of jump distance
can be obtained (inset).
Temperature Dependence of Segmental Relaxation
The characteristic relaxation time of the segmental process, , demonstrates strongly non non-Arrhenius
Arrhenius
temperature dependence in all polymers. It is usually described using Williams-Landel-Ferry (WLF)
C T T *
log aT log (T *) / (T ) 1
equation:
T T * C
Here T* is some reference temperature, often T*~Tg.
2

The same temperature dependence in non-polymeric systems is usually described by the Vogel-
Fulcher-Tammann (VFT) equation: B The equations are equivalent: C2=T*-T0;
0 exp
T T0 2.3C1C2=B, C1(T*)=log[(T*)/0].

Different polymers show stronger or weaker deviations of


from the Arrhenius behavior. The classification on strong and
fragile systems has been suggested by Angell [in: Relaxation in
Complex Systems, NRL, Washington, Eds.K.Ngai, G.Wright, 1984, p.3]. The
systems that show Arrhenius-like temperature dependence of
STRONG were called strong, strongly non-Arrhenius fragile.

Degree of fragility can also be related to an apparent


activation energy Ea around Tg (just a slope of ln() vs 1/T
around T~Tg). Ea~B/(Tg-T0) for some polymers can be large
than the binding energy for C-C bond and has therefore no
physical or chemical meaning. It reflects some cooperativity
FRAGILE
in motion.
of segmental relaxation depends on
molecular weight of a chain: the lower is
Mw the shorter is . This effect might
depend on end groups of the chain.
Traditional explanation is based on free
volume ideas (will be discussed later).
Relaxation time in PDMS with different
degree of polymerization (shown by
numbers on the left) [Roland, Ngai,
8 280
Macromolecules 29, 5747 (1996).]. In the case of
PDMS the temperature variations of
PDMS,
for all molecular weights scale well with
Tg (right) suggesting no change in
fragility.

Similar results are presented for PS


[Santangelo, Roland, Macromolecules 31, 4581
(1998)]. However, the fragility of PS chain
appears tot be
b dependent
d d t on Mw.
M

Recent ideas relate fragility to a rigidity of


backbone and side groups. Rigid chains
f t t packing
frustrate ki andd increase
i f ilit
fragility.
Thermorheological complexity (breakdown of time-temperature equivalence)
Developments of experimental techniques that are able to cover more than 5 orders in frequency provide
more accurate tests of time-temperature equivalence. They clearly demonstrate breakdown of time-
temperature equivalence principle for most of polymers: it appears that the shift factor is different for
different relaxation modes.

Atactic polypropylene, shift factors (right) and relaxation


times (left) for terminal and segmental relaxations
[Macromolecules 29, 3651 (1996) and Macromolecules 34, 6159 (2001)].

It seems that different processes have similar temperature dependence at high temperatures only
[Roland, et al. Macromolecules 34, 6159 (2001)].
Breakdown of time-temperature equivalence
appears also in the temperature dependence of the
rubbery plateau in entangled polymers.

Compliance data for PVAc obtained at two


reference temperatures (35 C and 60 C) show
shortening of the plateau region by nearly two
orders [Plazek, Polym.J. 12, 43 (1980)].

Dielectric relaxation spectroscopy allows


Mw~810 measurements of normal and segmental modes
at the same temperature, without use of time-
Mw~2000 temperature
p superposition.
p p Normal modes
show weaker temperature dependence than
Mw~4000
segmental relaxation. Data for PPG with
different Mw [Dielectric spectroscopy of Polymeric
Materials, ed. J.P.Runt and J.J.Fitzgerald]].
Comparison of normal mode relaxation time to the segmental
shows that theyy have similar temperature p dependence
p at higher
g T.
A difference increases drastically when the polymer approaches
the glass transition temperature Tg [Dielectric spectroscopy of Polymeric
Materials, ed. J.P.Runt and J.J.Fitzgerald].

Change of stretching parameter with temperature, different


temperature dependence for - and - processes are additional
evidences of the breakdown of time-temperature equivalence.

The reason for the breakdown of the time-temperature equivalence remains the subject of discussions.
First of all, segmental relaxation is driven by energetic forces, while Rouse modes are entropic and
reptation is controlled by a disenteglement time.
time Temperature dependence in the Rouse and reptation
models is introduced through monomeric friction coefficient. Temperature variations of the friction
coefficient may be different from temperature dependence of segmental relaxation.
Reptation model assumes that the same friction coefficient is involved in the Rouse modes at short
times (inside the tube) and in the reptation modes at long times. The data on PVAc show that the
friction coefficient might be different. Ferry and co-workers proposed [J.Colloid Sci. 14, 135(1959) and 17, 10
(1962)] that a number of entanglements can be a function of temperature. It is not clear whether this is
really the case.

Thus, one should be aware that time-temperature equivalence (superposition), although very often used
and presented in all textbooks, is oversimplification that fails for many polymers.
Concluding Remarks:

1. Time-temperature superposition assumes that all relaxation processes have similar temperature
dependence. It seems to be correct at high T, but it fails when temperature approaches Tg.

2. Spectral
p shape
p of segmental
g relaxation is always
y stretched ((non-exponential).
p ) Stretchingg of the
relaxation spectrum is a characteristic feature for relaxations in complex systems. The
mechanism of the stretching (dynamic heterogeneity or intrinsically non-exponential process)
remains unclear, although there are many indications of heterogeneous dynamics at time scale
smaller than segmental
g relaxation time.

3. Segmental relaxation exhibits non-Arrhenius temperature dependence that is usually


approximated by WLF or VFT equations. Steepness of temperature dependence of segmental
dynamics
y ((fragility)
g y) depends
p on chain rigidity.
g y
Glass Transition
Phenomenon of the Glass Transition
Glass transition is usually defined as a transition from a liquid state to a solid state. It appears as a sharp
change of temperature dependence for many properties, including volume (density), entropy, elastic
constants. The temperature where the change happens is called the glass transition temperature Tg.

Temperature variations of volume. Tg depends on Change in specific heat at temperatures around Tg.
cooling rate.

However, Tg is an ill-defined quantity. It depends on a cooling or heating rate. Due to that reason also
another definition of Tg is accepted: Tg is a temperature where segmental relaxation time ~100 sec.
There is no phase transition of any kind at Tg.

The nature of the glass transition phenomenon remains a subject of discussions. However, the basic
event of the observed transition from a liquid to a solid is a kinetic phenomenon. Glass transition is a
freezing of segmental relaxation.
Glass transition depends on pressure: increase in
P leads to increase in Tg of polymers. dTg/dP is
diff
different
t for
f different
diff t polymers.
l

This effect might be important for polymer


processing where polymer melt is usually under
some pressure.

The Kauzmann paradox


Considering changes of entropy, S, during cooling of a
liquid state,
state Kauzmann paid attention [Kauzmann,
K Ch
Chem.Rev.
R 43,43
219 (1948)] that extrapolated S of supercooled liquids may
become lower than entropy of a crystal. That should
happen at some temperature TK.
That does not violate any thermodynamic law.law However,
However it
is difficult to expect that the entropy of a disordered state
will be below the entropy of an ordered state.
TK has been found close to T0 VFT. That leads to
speculation on existence of a real thermodynamic transition
at T~TK~T0, that is avoided because the system falls out of
equilibrium at Tg.
Several models were proposed to explain VFT or WLF equations for and to describe the glass
transition:
-The free volume approach, assumes that the fractional free volume becomes 0 at T~T0;
-Thermodynamic approach (Adam and Gibbs theory, Gibbs DiMarzio theory) treats the glass
transition as a cooperative process, the degree of cooperativity increases when temperature
decreases.
However, all the models are phenomenological, have some problems and the nature of this
temperature dependence remains unclear.

Free Volume Approach


The basis is the Doolittles viscosity equation: a exp(bv / v f ) (1)
Here vf is the free volume and v is the total volume. The same relation can be written for the relaxation
time. The free volume is defined as vf=v-v0. Here v is the total macroscopic volume and v0 is the actual
molecular volume. The fractional free volume is usually assumed to vary with T:
f v f / v f g f (T Tg ) (2)
fg is the fractional free volume frozen at Tg and af is the thermal expansion coefficient of the free
volume Substituting eq.2
volume. eq 2 in eq.1:
eq 1:
b 1 1 b T Tg
log aT log
g 2.3 f f g
(3)
2.3 f g f g / f T Tg

Thus we have an equation similar to WLF equation, with T0=Tg, Tg, C1=b/2.3f
b/2.3fg; C2=ffg/af. In that case one
can relate the model parameters and WLF parameters: fg=b/2.3C1 and af=b/2.3C1C2. However, because
the constant b is an arbitrary parameter, no direct estimates of the free volume fraction can be obtained.
f A simple picture behind the free volume approach is based on the
assumption that an empty (free) volume is needed for molecular
V motion.
ti Thi free
This f volume
l d
decreases with
ith decrease
d i T.
in T That
Th t leads
l d
Vg fg Vo to slow down of the motion. Relaxation time at Tg crosses the
Vg experimental time scale and freezing of the structure (including
free volume) occurs. It gives frozen free volume fg. If one would
T Tg T cooll the
th sample l down
d with
ith infinitely
i fi it l slow
l rate,
t fg=0
f 0 willill be
b
reached at T that would be equivalent to T0 of VFT equation.
When one considers b~1, the value for the free-volume fraction at Tg for many polymers falls in the
range
g fg~0.013-0.034. WLF pproposed
p a universal value fg~0.025. Later,, another relation was also
suggested: fg=10-4*Tg+0.07 [Boyer, Simha, J.Polym.Sci.Polym.Lett. 11, 33 (1973)]. It was further modified: fg is not
frozen at Tg; one should distinguish fractional empty free volume or dynamic free volume from total
free volume.
Free volume approach has been extended to include pressure effects:
b T T0 ( P )
log aT (4)
2.3 f 0 f 0 / f T Tg ( P )
Here f0 is f at T=T0. (P) is a function that depends on the pressure-dependent coefficient of thermal
expansion of the free volume, f(P). It has been shown that the equation 4 describes reasonably well T-
and P- dependencies for some polymers.
Nevertheless, the free-volume approach has been criticized for many problems. In some cases
unreasonable parameters of the free volume should be assumed in order to describe data for some
materials. It has been also demonstrated that holding free volume constant (by varying simultaneously P
and T) leads to different viscosity, suggesting that not only density, but also temperature play role in the
glass transition.
The influence of molecular weight on Tg, chain-end free volume approach
Tg for many polymers depends on molecular weight Mn. At not very low Mn, Fox-Flory empirical
equation describes reasonably well the molecular weight dependence of Tg:
K
Tg ( M n ) Tg () (5)
Mn
pp
Free volume approach explains
p the Fox-Floryy equation,
q assumingg that chain ends contribute an excess
free volume. In that case, decrease in Mn leads to increase of chain ends concentration and increase of
free volume. Increase in free volume leads to decrease in Tg.
Lets assume that chain end has a free volume . Then the free volume per unit volume is 2NA/Mn,
where is density,
y NA is Avogadros
g constant. Assumingg that fgg is independent
p of Mn, the excess free
volume introduced by chain ends should be compensated by the thermal contraction:
2N A
2N A / M n f [Tg () Tg ( M n )] Then Tg ( M n ) Tg () (6)
f Mn
The Fox-Floryy equation
q A/fMn.
is obtained with the constant K=2N
380

360 Thus, chain-end free volume idea describes well the


Tg(Mn) in PS molecular weight dependence of Tg. However, it is
Tg(Mn) [K]

340

320 known that the free volume is not universal at Tg.


Tg
300
Also, a simple free volume approach can not explain
280
the behavior of Tg in ring polymers (will be
260
Tg(inf)-K/Mn discussed later.
240

220
3 4 5 6
10 10 10 10
Mn
Thermodynamic Approach
Any conformational
A f ti l changes
h require
i some cooperative
ti motion
ti off
a few molecules (cooperative domain). Domain consist of z
conformers, each has c1 number of states. The conformational
entropy for 1 mole of conformers in which there are Nz domains
consisting of z conformers: S c N z k ln c1
N A N A k ln c1 s *
z
Nz Sc Sc
here ss* is the conformational entropy of one mole of conformers
where each conformer relaxes independently. At high T>T* when
there is no cooperativity s*=Sc. At T<T* Sc drops faster than s*,
assumption
T * T T0 C
Sc s*; 0 exp
T * T0 T c
TS

An equation
q is the energy
for relaxation time (( gy barrier for one conformer to relax):
)

z 1 s * 1 T T *
ln / *
k T T * k TS c T *
(7)
kT * T T0

The eq.7 is equivalent to WLF or VFT equations.


Gibbs DiMarzio theory

The theory is based on application of the lattice model of a polymer


system to the glass transition problem. The polymer chains of degree
of polymerization X have many configurations which fit onto the
lattice of coordination number Z. Each chain has the lowest energy
shape. Deviation of each bond from the lowest energy shape cost
energy . The number of flexed out bonds is f. There are also n0
vacant sites (holes) on the lattice. It results in additional hole energy
per intermolecular bonds broken by introduction of the vacancies into
the lattice (bond energy ).

Thee lattice
a ce modelode ppredicts
ed c s thee eexistence
s e ce oof a true
ue seco
second-order
d o de
transition at a temperature T2. The number of allowed arrangements of
the molecules decreases with decreasing T because: (i) the number of
holes decreases; (ii) the configurational entropy of the molecules
deceases because the chains favor low-energy gy states at lower T.The
T(P) transition line defines the point T2(P) where the total
configurational entropy first becomes 0. In that respect, the T(P) line
represents the thermodynamic glass transition in experiments of long
time-scale.
Gibbs DiMarzio theory gives rather complicated prediction for the molecular weight dependence of
Tg
g [McKenna, Compreh.Polym.Sci. 2,311 (1989)]]:
2 exp
x ln v0 1 v0 ( x 1)(1 v0 ) ln 3( x 1) kTg kT
ln 1 g
ln 1 2 exp (8)
x 3 1 v0 1 v 0
1 2 exp 2kTg
2 xv0 x
kT
g

Here x is twice the degree of polymerization and v0 is the volume fraction of holes, Tg is the glass
transition temperature T2. The eq.8 describes well Tg(Mn). It even describes the deviation from the Fox-
Flory equation at smaller Mn.
The theory also makes an interesting prediction for ring polymers. It has been observed that Tg increases
with decrease in Mn in ring PDMS. This behavior has been described qualitatively using the eq.8.

PVC PDMS

rings

linear

Tg vs 1/M in PVC. Solid line shows Fox-Flory relationship. Tg in ring and linear PDMS compared to predictions
The dashed lines shows Gibbs DiMarzio model predictions of Gibbs-DiMarzio theory [from Guttman, DiMarzio,
(eq.8) [from Pezzin, et al. Eur.Polym.J. 6, 1053(1970)]. Macromolecules, 1988]
Influence of Molecular Structure and Architecture on Tg
Flexibility of the main-chain. Flexible group (for example, ether link) will enhance main-chain
flexibility and reduce Tg. Inflexible group (for example, terephthalate) will increase Tg.
Si-O-Si is a very flexible link. As a result, PDMS has one of the lowest Tg known for polymers.

Influence of side groups. Bulky, inflexible side groups increase Tg .


Rigid and flexible side groups:

Increasing the length of side groups reduces Tg, mostly due to increase in free volume.
Role of tacticity and microstructure. Tacticity of polymer chain can affect flexibility of the chain,
packing of chains (free volume) and in this way will influence glass transition temperature. Tacticity
strongly
l affects
ff crystallization
lli i off polymers.
l

Example of PMMA: Tacticity in PMMA influences


significantly Tg and its
syndiotactic isotactic dependence on M. Data from
OCH3 H 3C H3C [Ute, et al. Polymer 36, 1415
CO CH3 (1995)].
C C C
C C C C C
C C
CO CO
H3C CO
OCH3 OCH3
OCH3
Microstructure of polymers also affect Tg significantly. An example is 1,4 and 1,2 poly(butadiene).
Presence
ese ce oof thee doub
doublee bo
bond
d in thee bac
back-bone
bo e reduces
educes Tgg ddrastically.
as ca y.
C C
C C 1,4-PB, Tg~175K

C C
C C 1,2-PB, Tg~270K
-C C-
Tg in polybutadienes is a smooth function of vinyl content [Hofmann,
H f
et al. Macromolecules 29, 129 (1996)]. This is an example of Tg in a
random co-polymer.
150

350 It is important
p to emphasize
p

K]
Tg [K
140
Me that the molecular weight
Tg [K]

300 Me dependence of Tg does not


130
correlate with molecular
weight
g between
PDMS 250 PS entanglements, Me.
120
1000 10000 3 4 5 6
Mn 10 10 Mn 10 10
PDMS and PS have similar Me but differ strongly in the molecular weight dependence of Tg.

Influence of end groups. We already discussed the idea of


chain-end free volume approach that explains the molecular
weight dependence of Tg. We also discussed that ring polymers
(no ends) have different dependence of Tg on Mn.
It is possible to change direction of the molecular weight
dependence of Tg by changing end groups. The stronger are
interactions of the end groups the stronger will be increase of
Tg with decrease in Mn.

Dependence of Tg on chain length (xM number of monomers)


in PFB polymer with different end groups [Danusso,
Danusso et al.
al Polymer
34, 3687(1993)]: - CH2OCH3 ; - CH2OSi(CH3)3
o CH2OCOCF3 ; - CH2OH ; + - CH2OK.
Influence of molecular architecture. Following chain-end free volume ideas (eqs.5,6), some authors
[Roovers, Toporowski, J.Appl.Pol.Sci. 18, 1685 (1974); K.L.Wooley, et al., Macromolecules 26, 1514 (1993)] proposed
Tg Tg
modified
difi d Fox-Flory
F Fl equation
i forf star polymers:
l f K
2 Mn
Here f is the number of arms and Mn is the total molecular weight. In that case, a linear chain is
considered to be a two
two-arms
arms star.

The proposed dependence has been observed for PS stars


with 4 and 6 arms [Roovers, Toporowski, J.Appl.Pol.Sci. 18, 1685
(1974)].

This dependence, however, has not been observed in PI


stars [C.Kow, et al., Rub.Chem.Tech. 55, 245(1982)].

180 180

178 178 Analysis of PB stars demonstrate that Tg depends


Tg [K]

more on the total Mn than on architecture [A.


A Kisliuk,
Ki li k ett
176 176 al., J.Polym.Sci.B 40, 2431-2439 (2002)].

174 174
Linear Thus influence of molecular architecture on Tg and
Star segmental dynamics remains unclear and
172 172
0 1 2 3 4 5 0 2 4 6 8 10 experimental data are controversial.
4 4
10 /Mn,tot 10 f/Mn,tot
320
Validity of the chain-end free volume
approach
pp even for dendritic ppolymers
y was
proposed in [Wooley, et al. Macromolecules 26, 1514 300
(1993)]. Tg in dendritic poly(benzyl ethers)

Tg [K]
with various microstructures scales with 1/M.
However, it appears
pp that it scales better with 280
Mtot than with f/Mtot.
[G]-OH
260
[G]2-[C]
[G]3-[C]

0 5 10 15 20 25 30 50 60 70 80 90
4 4
10 /Mtot 10 f/Mtot

Influence of crosslinking. Chemical cross-links increase Tg.


The main reason is an additional restriction on molecular
motion,
ti th t reduces
that d segmental
t l mobility.
bilit At high
hi h
concentration of cross-links, where motions of segments is
significantly restricted, there is no glass transition.
Variation of Tg in polymer solutions, co-polymers and blends. Glass PVC/solvent PVC-100%
transition depends strongly on composition of polymer solutions, on
a solvent used. Adding a solvent to a polymer usually leads to
plasticization (decrease of Tg). The figure shows PVC plasticized
with various amounts of di(ethylhexyl)phthalate.
Tg of PS solutions as a function of solvent
concentration in different
ff solvents.

-naphthyl salicylate
phenyl salicialte
tricresyl
phosphate Tg varies non-linear with concentration of solvents. One of the
methyl explanation is based on the free-volume approach. Fractional free
acetate volume of the solvent fs is higher than that of the polymer fp (solvent
benzene has lower Tg). Assuming that the fractional free volumes are not
nitrobenzene additive, the total fractional free volume:

f vs f s v p f p k v vs v p (11)
Here vs and vp are volume fractions, k is a negative constant ~10-2.
Eq.11 assumes that occupied volumes are additive and that agrees
with measurements of density. Assuming iso-free-volume state at Tg
the following dependence is predicted [McKenna]:
vs sTgs v p pTgp k v vs v p (12)
Tg ( vs )
vs s v p p
Here is the thermal expansion coefficient of the fractional free volume. The eq.12 describes the data
well with reasonable parameters.
Free volume ideas were also used for description of Tg in co-
polymers. Additivity of volumes of the two components is assumed Poly(butadiene-co-styrene)
] T Tg vR w1vR1 w2 vR 2
[Gordon, Taylor, J.Appl.Chem. 2, 493 (1952)]:
T Tg vG w1vG1 w2 vG 2
Here vR and vG are the specific volumes of the copolymers in the rubbery
and glassy states,
states w is the weight fraction.
fraction The authors predict for Tg:
kw T w2Tg 2
Tg ( w2 ) 1 g 1 (13)
w1 kw2
Here k=2/1, is a difference of the thermal expansion coefficient
in the rubbery and in the glassy state. The eq. 13 describes the data for
poly(butadiene-co-styrene) reasonably well. Note that the eq.13 is
identical to the eq.(12) with kv=0.

Similar ideas have been used to describe Tg in block co-polymers.

In the case of ppolymer


y blends there are two situations:

Immiscible blends show separate Tg for each of the components (2 Tgs


for binary blends). Example of shear modulus and logarithmic
decrement in blend of PS and styrene-butadiene
y co-polymer.
p y
Separated Tgs will be observed also in block co-polymers of
immiscible blocks.
A single Tg appears in DSC of
miscible blends. It appears in
b t
between T off the
Tgs th mixed
i d
components and the transition
broadens. However, each
component has its own relaxation
ti
time, andd off the
th fast
f t componentt
slows down less than of the
slow component speeds up. One
can analyze it as Tgeff.

There are a few models proposed for the description of Tg in miscible polymer blends. One of the recent
ideas proposed by Lodge and McLeish [Macromolecules 33, 5278 (2000)] takes into account effective
concentration of monomers A and B around a typical monomer A: eff S (1 S )
Where is the volume fraction of A component and S is the self-
C m
concentration term: S 0 (14)
n0 N AV
Here m0 is mass of a monomer, n0 is the number of backbone bonds per
monomer, is the density, V is a volume that influences the relaxation of
a monomer. It is assumed to be V~lK3 (lK is the Kuhn length).The main
idea is that the monomer A experience higher concentration of A because
of its connectivity. As a result, you have a distribution of eff. The model
explains many characteristic properties of relaxation and Tg in miscible
blends, even gives good quantitative predictions for some systems.
However, it fails for some other systems.
Glass transition in thin polymer films: Developments in nano science and nano technology
leads to decrease of characteristic size of elements. Qualitative difference appears when we
approach length scales ~5-50 nm.
One example of that is variation of Tg in thin polymer films [review by Forrest, Dalnoki-Veress, Adv.Coll.Interf.Sci.
94, 167 (2001)]. It has been observed that Tg of polymer films drops with decrease in film thickness h. The
best ppolymer
y analyzed
y is PS. Various techniques
q show dropp of Tgg ~30-50K when h ~10 nm. Dropp of Tgg
depends on substrate, suggesting importance of a polymer-substrate interactions. In particular, it has been
shown that for polymer physically grafted to substrate Tg can increase with decrease in h.
Another example: PMMA on Au -> Tg decreases with decrease in h, while PMMA on SiO2 -> Tg
increases with decrease in h.

Tg in thin PVAc films [Fukao, et al., J.Non.Cryst.Sol. 307-


Tg in thin PS films as a function of thickness h. 310, 517 (2002)].

No significant dependence of the effect on Mw has been observed in thin supported films.
films Tg(h) is

usually described by an equation: a
Tg ( h ) Tg 1
bulk

h
Even stronger effect has been reported for thin free
standing films (no substrate). In that case, strong
d
dependence
d off the
th effect
ff t on molecular
l l weight
i ht has
h
been observed. There are two regimes: lower Mw
(for PS~120,000 370,000), and higher Mw
(>370,000).
Eff t att lower
Effect l Mw is i similar
i il tot the
th effect
ff t in i
supported films but is ~2 time stronger.
Much stronger dependence appears at higher Mw, it
starts to deviate from the bulk behavior at h~REE
( dt
(end-to-endd distance).
di t )

Traditional picture for these variations is related to a surface layer that is assumed to have higher
mobility than the bulk polymer.
polymer Of course,
course if it interacts strongly with a substrate,
substrate it might have lower
mobility. The thickness of the layer might be different for different polymers and for PS is estimated to
be ~3.5-4 nm. Decreasing the thickness of the film enhances the influence of the layer on the film
properties.
Surface layer
y
Bulk properties Film
Surface layer

This idea explains why the effect appears to be stronger in free-standing


free standing films.
films There are many
microscopic models that try to explain details of the effect, in particular, for free-standing films with
high Mw polymers. However, no one of them can describe all the data consistently.
Concluding remarks

1. Glass transition in polymers is freezing of segmental relaxation. Tg is an ill-defined


quantity. It corresponds to the temperature at which segmental relaxation time becomes
comparable to the laboratory time scale (cooling rate).
rate)

2. There are two main approaches for description of the glass transition: free volume and
thermodynamic (Gibbs-DiMarzio). Both describe the phenomena qualitatively and both have
various problems.
problems

3. Chemical structure, chain length and architecture of macromolecules affect their glass
transition temperature.

4. Confinement effect leads to variation of Tg in thin polymer films. Detailed microscopic


mechanism of these variations remains unclear.

Anda mungkin juga menyukai